Identification of irregularities and allocation suggestion of relative file system permissions

This paper presents the novel use of statistical analysis to establish independence and homogeneity in allocated file system permissions. This independence can be interpreted as potential anomalies in a system's implementation of access control. The paper then presents the use of instance-based learning to suggest the allocation of new permissions conforming to a system's current implementation structure. Following this, both of the presented techniques are then included in a tool for interacting with Microsoft's New Technology File System (NTFS) permissions. This involves experimental analysis on six different NTFS directory structures within different organisations. The effectiveness of the developed technique is then established through analysing the true positive and true negative values. The presented results demonstrate the potential of the proposed techniques for overcoming complexities with real-world file system administration.